Wrestle AI
Wrestle AI is an iPhone app that coaches wrestling technique. A wrestler uploads a clip of a practice or a match, and the app returns a score, strengths and weaknesses, drills that fix the mistakes, and strategy notes. It was built with Rork, a no-code AI app builder: a service that writes and publishes a phone app from instructions in plain language, so the maker writes no program code. The analysis runs on OpenAI models. Running a trained model on new input, here the frames of a video, is called inference, and it is the cost that grows with every video. Every feature sits behind a paywall, a screen that asks for payment before the app can be used: US$9.99 a month or US$59.99 a year.
George Lampropoulos, an 18-year-old college student in the United States who says he does not know how to code, found Rork in June 2025. The first App Store version is dated 18 September 2025. Rork's case study of 25 February 2026 says the app earned US$131,863 in total revenue, counted from August 2025, with MRR (monthly recurring revenue, the subscription income of one month) of US$23,000, and US$39,400 of revenue in the month when yearly payments are counted in full. He spent about US$1,100 in total, including about US$250 for a freelancer. Users came through short videos by wrestling creators, whom he contacted with about 100 messages a day and paid per view. A large wrestling creator became his partner, whom Rork calls his co-founder; the podcast of Superwall, a company that makes paywall software for apps, lists a 50/50 partnership with the wrestling creator Cayden Henschel. Rork's press release put the app at over US$30,000 a month in April 2026.
Every money figure on this page is grade C. It comes from Rork, which sells the builder and uses this story in its marketing and in its funding announcement, from the founder, or from sites that repeat them. No payment processor or app-revenue platform is quoted, and the figures use different definitions: US$17,000 a month in December 2025, US$23,000 MRR in February 2026, over US$30,000 a month in April 2026. The latest figure, almost US$41,000 MRR in October 2026, covers three of his apps together. The page also shows earlier companies that promised feedback on how athletes move, and the risks of an app for teenage athletes in one sport.
The numbers, graded
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Total revenue to February 2026 | US$131,863 | C | Rork's case study, 25 February 2026, captioned "Aug 2025 to Feb 2026". Rork sells the app builder and does not say which payment system the figure comes from. Rork's post on X, quoted by AI Dev Cases, gives US$131,000 in about six months. | [1, 4] |
| Monthly recurring revenue (MRR), February 2026 | US$23,000 | C | Rork's case study. It does not say how yearly plans are counted in this figure. | [1] |
| Revenue in the month, February 2026, with yearly payments counted in full | US$39,400 | C | Rork's case study: "total monthly revenue hitting $39.4K when you include yearly subscriptions". It is about 1.7 times the MRR because the free trial is offered only on the yearly plan. | [1] |
| Run rate from the February 2026 MRR | about US$276,000 a year | calc | Calculated: US$23,000 times 12. | [1] |
| Average revenue per month to February 2026 | about US$21,977 over six months | calc | Calculated: US$131,863 divided by 6, Rork's window from August 2025. Counted from the first App Store version in September 2025, five months, the average is about US$26,373. | [1, 2] |
| Revenue per month, December 2025 | US$17,000, of which about US$8,000 from monthly subscribers | C | A December 2025 article on thestartupstorys.com, as summarised by AI Dev Cases; the article is no longer online. The other US$9,000 was yearly-plan revenue spread over the months. | [6, 4] |
| Revenue per month, April 2026 | over US$30,000 | C | Rork's press release on its US$15 million seed round, 9 April 2026: the founder "scaled it to over $30,000 in monthly revenue". | [5] |
| Revenue of his three apps together, October 2026 | almost US$41,000 MRR, and US$84,000 in 28 days | C | The Superwall Podcast episode of 15 October 2026: Wrestle AI, Brawler and a fitness app. No split per app is published, so this is not a figure for Wrestle AI alone. | [3] |
| Total spending to February 2026 | about US$1,100 | C | Rork's case study. Rork's post on X, quoted by AI Dev Cases: he "spent only ~$1,100 to reach $20K/mo". | [1, 4] |
| Running cost | under US$150 a month | C | Rork's case study: Rork for the app, Supabase for the backend, OpenAI for AI inference. | [1] |
| Outside help | about US$250 for a freelancer | C | Hired on Fiverr to add payments and sign-in, which Rork did not handle well at the time, according to Rork. | [1] |
| Downloads | more than 17,000 by February 2026, more than 100,000 by October 2026 | C | Rork's case study, then the podcast description: "Wrestle AI passed 100K downloads on creators alone". | [1, 3] |
| App Store ratings, 16 October 2026 | about 2,200 ratings, average 4.8 of 5 | B | Apple's count on the App Store page. Rork's case study showed about 1,700 ratings in February 2026. | [2, 1] |
| First App Store version | 18 September 2025 | B | Version 1.0 in the App Store version history. Rork says the app reached place 18 in its App Store category on launch day from pre-orders. Rork's revenue window starts in August 2025; the sources do not explain the difference. | [2, 1] |
| Seller on the App Store, October 2026 | WRESTLE AI LLC | B | The App Store page. Fight AI, his app for MMA, is sold by IMPROVESCAPE LLC together with five other apps. AI Dev Cases named IMPROVESCAPE LLC as the developer of Wrestle AI in August 2026. | [2, 8, 7, 4] |
| Team | the founder and one creator partner, plus a freelancer | C | Rork calls the partner "his co-founder - one of the largest wrestling creators online". The podcast description lists a 50/50 partnership with the wrestling creator Cayden Henschel. No source says whether 50/50 means a share of the company or of the revenue. | [1, 3] |
| Outside funding | none reported | C | No source mentions investors in Wrestle AI. The US$15 million seed round of April 2026 was raised by Rork, the builder. | [1, 5] |
| Time from the first App Store version to the February 2026 figure | 5 months | calc | Calculated: September 2025 to February 2026. Rork counts six months from August 2025. | [1, 2] |
How to read this table: a grade C value is a statement by Rork, the company that sells the app builder, by the founder, or by a site that compiles their statements. A grade B value is a record that Apple shows on the App Store. A "calc" value was computed on this page from the rows above it. No money value is grade A or B.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Basic Monthly | 9.99 | all features | |
| Basic Yearly | 5.00 | US$59.99 a year; 3-day free trial | |
| Discount Yearly | 2.50 | US$29.99 a year; offer price |
Every feature needs a subscription. Rork's case study of February 2026 says the yearly plan has a 3-day free trial and the monthly plan has none, which pushes buyers to the yearly plan. The App Store also lists a Discount Yearly plan at US$29.99, an offer price. The yearly plans are paid once a year; the per-month values in the table are the yearly price divided by 12.
The timeline
From a failed app built by agencies, started at age 15, to an app built alone with prompts and sold through creators. Each entry carries its grade.
- Age 15CWith a friend he builds an audience of 10,000 followers for a social self-improvement app, then pays three development agencies to build it. It takes a year and a half; by launch the audience has gone and the savings are spent. [1]
- June 2025CHe finds Rork and builds Wrestle AI with prompts, from morning to night, pasting error logs into ChatGPT when something breaks. A freelancer from Fiverr adds payments and sign-in for about US$250. [1]
- 18 September 2025CVersion 1.0 on the App Store. Rork says the app reached place 18 in its category on launch day from pre-orders. [2, 1]
- 29 September 2025CAnswering a review about the price, the developer writes that the app processes hundreds to thousands of frames per video and pays very high AI inference costs. [2]
- 13 October 2025CVersion 2.0 adds a practice mode, a match-day poster generator and a weight-cut mode. [2]
- 17 November 2025CVersion 3.0 adds a new calorie tracker, an opponent analysis mode, a journal and a weight management dashboard. [2]
- December 2025CA feature article reports US$17,000 a month, about US$8,000 of it from monthly subscribers. [6, 4]
- 25 February 2026CRork's case study: US$131,863 in total revenue, US$23,000 MRR, about US$1,100 spent, more than 17,000 downloads. The founder runs weekly marketing sessions on Rork's Discord server and wrote the marketing guide in Rork's documentation. [1]
- 9 April 2026CRork announces a US$15 million seed round and names Wrestle AI as an example at over US$30,000 a month. [5]
- 13 April 2026CVersion 5.4 adds Brazilian jiu-jitsu, freestyle and Greco-Roman wrestling. [2]
- 14 April 2026COn X, the founder writes that his published method led to hundreds of copycat apps for other sports and that almost none of them made money, as quoted by AI Dev Cases. [4]
- 28 July 2026CHe publishes a Growth Playbook about the apps that took him "from $0 to $50k a month", as quoted by AI Dev Cases. The figure covers all his apps. [4]
- 4 September 2026CVersion 5.5, called Wrestle AI 2.0, adds a social area, The Locker Room, and a new interface. [2]
- 15 October 2026CThe Superwall Podcast: his three apps make almost US$41,000 MRR and Wrestle AI passed 100,000 downloads from creators alone. The episode description lists the 50/50 partnership with the wrestling creator Cayden Henschel. [3, 9]
- 16 October 2026BThe App Store lists WRESTLE AI LLC as the seller, version 5.9 of 1 October 2026, and about 2,200 ratings with an average of 4.8. [2, 8]
Why it worked
Four causes, in the order they mattered. Each one is a reusable lesson.
A sport he had practiced
He wrestled through middle and high school, and he knew that wrestlers rarely have an expert to review their film. The first version did one thing: upload a clip, get a scored breakdown. Rork's post on X gives the reason that so little money was needed: there were no other apps like it on the App Store.
The feature is the advertisement
The founder calls it the "gotcha moment": the raw match clip, then the AI breakdown with drills, which can be shown in under 10 seconds. Creator videos that look like a day of training, with the app in the middle, sold it better than polished ads. The podcast description says a raw screen recording reached a cost per acquisition (CPA), the advertising cost of one new user, of US$10, while polished ads failed.
Creators paid per view
He sent about 100 direct messages a day to creators, opened with "Paid promo?", and negotiated by phone, never by text. He paid US$2 to US$5 per thousand views, a price called CPM, with a guaranteed number of views and a cap on the total. One of these contacts, a large wrestling creator, became his partner. The podcast says one US$4,500 creator deal paid back 75% of its cost, about US$3,375.
Onboarding that ends at a paywall
New users answer questions on weight class, goals and position, upload a video and see a preview of the analysis; only then does the app ask for payment. The free trial exists only on the yearly plan, which pushes buyers to pay US$59.99 for a year at once. That is why revenue in February 2026 was US$39,400 in the month against US$23,000 of MRR.
Almost no fixed cost
Rork wrote the app, Supabase stores the data and accounts, and OpenAI models write the feedback. Spending to February 2026 was about US$1,100 in total and the running cost under US$150 a month. The cost that grows with use is the AI analysis of each video, which the developer called "very high" in a reply to a user in September 2025.
How it was built
These details come from Rork's case study, the App Store page, the podcast description and AI Dev Cases. They are grade C. [1, 2, 3, 4, 7, 8, 13]
- Built with Rork from prompts, with no code written by the founder. When something broke, he pasted the error log into ChatGPT and asked for the fix.
- Payments and sign-in were added by a freelancer hired on Fiverr for about US$250, because Rork handled them badly at the time.
- Supabase for the backend, which holds the database, the accounts and the stored files, and OpenAI models for the analysis and the written feedback. Running cost under US$150 a month.
- The analysis reads frames taken from the uploaded clip. A user review of January 2026 says that full matches must be trimmed first and that uploads repeatedly failed to analyze.
- 19 App Store versions from September 2025 to October 2026 added drill plans, a calorie tracker, a weight-cut mode, opponent analysis, Brazilian jiu-jitsu and a social area.
- The same method produced Fight AI, an app for MMA, which is sold by IMPROVESCAPE LLC. Wrestle AI is sold by a separate company, WRESTLE AI LLC.
- The privacy policy, terms of service and license agreement linked from the App Store page are hosted on Termly. The privacy policy could not be read for this page.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
The dead that came before
Four companies in the Loot Drop database promised feedback on how a person moves, or an app for one sport. The values below are from that database and are not independently verified.
Enflux
The promiseClothing with sensors that tracked movement in 3D and gave athletes real-time feedback on their form.
Why it diedLarger companies had more money and trusted brands. The niche product never sold in the volume needed to make it cheap, and production was expensive.
FitFu
The promiseA phone app that detected exercises with the phone's motion sensors and turned workouts into a game.
Why it diedUsers left after the novelty, it had no strong way to charge, and Nike Training Club and Fitbit offered more complete products.
Charge Running
The promiseLive online running classes with real-time audio coaching.
Why it diedIt could not win and keep users against larger fitness apps. Classes at fixed times lost appeal, and on-demand AI coaching undercut it.
Rally Tennis
The promiseAn app that matched amateur tennis players with local opponents of the same level.
Why it diedIt needed many active players in each city, and users would not pay enough to keep it running.
What changed
Enflux put sensors in clothing and FitFu used the phone's motion sensors to judge an exercise. The records say that the clothing was expensive to make and that the phones of the time were too limited for better features. Charge Running and Rally Tennis were apps for one activity that could not win or keep enough users: Charge Running against larger fitness apps, Rally Tennis because it needed many players in each city. All four spent investors' money, from US$120,000 to US$5 million.
Wrestle AI needs only an ordinary phone video and a general AI model, so it has no hardware cost and is useful to one user alone. It was built for about US$1,100 in total, so it reached revenue without investors, and its users came from creators who already had the audience. Two things did not change: it is an app for one sport, like Rally Tennis, and it competes with large team-video companies such as Hudl for the same athletes.
Can one person copy it now?
The method, yes. The empty niche, only once per niche.
The tools are open to anyone: a no-code AI builder, a general AI model for video, a freelancer for payments, and creators paid per view. The founder writes that hundreds of copycat apps for other sports followed his published method and that almost none made money. Rork gives the reason as being first in a sport that had no such app, and the founder names his own persistence and his choice of niche.
Where it is still open: Sports where athletes film themselves and few coaches are available, as long as no app serves them yet. The founder's own warning applies: copies of his app for other sports mostly made no money.
- 1Pick a sport or skill that you have practiced yourself, where people film themselves but have no expert to review the film.
- 2Build one feature first: upload a clip, get a score and drills. Check that the result can be shown in a 10-second video.
- 3Send many short messages to creators in that niche and pay per view, with a guaranteed number of views and a cap.
- 4Put the paywall after the user has answered the onboarding questions and uploaded a video, and offer the free trial only on the yearly plan.
What to watch
The risks are the mirror image of the causes above.
The figures come from the tool maker
Every revenue figure comes from Rork, which sells the builder and uses this story in its marketing and in its funding press release, or from the founder, who runs weekly sessions on Rork's Discord server and wrote the marketing guide in Rork's documentation. No payment processor or app-revenue platform figure is public. The monthly figures range from US$17,000 to over US$30,000 depending on the date and the definition.
Large sports-video companies
Hudl, founded in 2006 and based in Lincoln, Nebraska, sells video and data tools to teams, from youth teams to professional clubs. It bought Ubersense, a slow-motion video app for athletes, in September 2014, and its competitors Krossover in May 2019 and Wyscout in August 2019; Bain Capital invested in May 2020. Its site says that its AI algorithms tag and track what teams film. Wrestling apps such as MatBoss and WrestlingIQ appear next to Wrestle AI in the App Store, and similar AI wrestling coach apps from another developer exist.
Videos of school-age athletes
The App Store page names youth and high school athletes as users, and the app is rated 13+ and has a weight-cut mode. In the United States, the COPPA law applies to online services that are directed to children under 13 or know that they collect data from them; courts can fine up to US$53,088 per violation, and in August 2026 TikTok and ByteDance agreed to pay US$400 million to settle a COPPA lawsuit. The app's privacy label lists usage data used to track users across other companies' apps and websites, and advertising data; it does not list the uploaded videos. The privacy policy on Termly could not be read.
One sport and one creator
Wrestle AI serves one sport, its users arrived through creators, and one wrestling creator is a 50/50 partner. No public source says what the partnership covers, who owns WRESTLE AI LLC, or what happens if the partner leaves. The October 2026 figure covers three apps with no split, so the trend of Wrestle AI alone after April 2026 is not known.
Dependence on Rork and OpenAI
Rork wrote the app and OpenAI models run the analysis. A price change at OpenAI changes the cost of every video, and the developer already called the inference costs "very high" in September 2025. A review of January 2026 says uploads repeatedly failed to analyze. A founder who does not code needs Rork or freelancers for every repair.
Other stories built without code
Several businesses on this site were built with Bubble, a no-code builder where the maker arranges parts on a screen: AudioPen, Formula Bot, Magai and My AskAI. No other story here names Rork, a builder that works from prompts, as its builder; the AppAlchemy story describes that market, where Rork raised US$15 million in 2026. The founder writes that hundreds of copies of his app for other sports made almost no money.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- C Rork blog, 25 February 2026: 18yo Made $131K in 6 Months with RorkRead in full. A case study by the company that sells the app builder. Source for the revenue, MRR, spending, costs, freelancer, tech stack, prices, marketing method and the earlier failed app. It does not name the system that the figures come from.
- B Wrestle AI on the US App Store, read 16 October 2026Read in full: description, prices, version history, privacy label, age rating, seller, ratings and reviews. Seller, dates and ratings are Apple's records; the description and prices are written by the company.
- C The Superwall Podcast, 15 October 2026: George Lampropoulos - Meet the ex-wrestler making $41k/mo with AI appsRead the episode page and description on Apple Podcasts. The 47-minute audio was not transcribed. Source for the three-app figure, the downloads, the 50/50 partnership and the advertising results.
- C AI Dev Cases, 24 August 2026: A Non-Coder, 18 — How Wrestle AI Turned a No-Code Build Into $131,863 in Six MonthsRead in full. A compilation that quotes the founder's and Rork's posts on X and summarises the December 2025 article. Source for those quotations and for the developer name in August 2026.
- C PR Newswire, 9 April 2026: Rork Raises $15M to Power the Next Generation of App Store EntrepreneursRead in full. Rork's own press release. Source for the April 2026 monthly figure and for Rork's funding.
- C thestartupstorys.com, December 2025: feature on Wrestle AI at US$17,000 a monthCould not open: the page returns 404, the site search does not find it, and the Wayback Machine has no copy. Its figures are cited from AI Dev Cases' summary.
- B App Store developer page: IMPROVESCAPE LLC, read 16 October 2026Read in full. Lists Fight AI and five other apps.
- B App Store developer page: WRESTLE AI LLC, read 16 October 2026Read in full. Lists only Wrestle AI.
- C The Superwall Podcast episode list, superwall.comRead the episode list. It lists the episode on 12 October 2026 with a link to the video.
- C Hudl: About page, read 16 October 2026Read in full. Source for Hudl's description of its video and AI analysis for teams.
- C Wikipedia: HudlRead the history and acquisitions sections. Source for the founding year, the acquisitions and the funding dates.
- C Wikipedia: Children's Online Privacy Protection ActRead the sections on scope, penalties and enforcement. Source for the age limit of 13, the fine per violation and the August 2026 TikTok settlement.
- C Wrestle AI privacy policy on Termly, linked from the App Store pageCould not read: the page loads its text by script, and Termly's interface answered "Policy Not Found" on 16 October 2026.
- C Loot Drop database export, 1,749 failed startupsLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.